The development of print tuning in children with dyslexia: Evidence from longitudinal ERP data supported by fMRI

Abstract

A consistent finding in functional brain imaging studies of reading with dyslexia is reduced inferior occipito-temporal activation linked to deviant processing of visual word forms. Time-sensitive event-related potentials (ERP) further revealed reduced inferior occipito-temporal N1 tuning for print in dyslexic 2nd graders suggesting the reduction affects fast processing and the initial development of dyslexia. Here, we followed up the same groups with ERP recordings and investigated how fast print tuning deficits in dyslexia develop from 2nd to 5th grade. Using functional magnetic resonance imaging (fMRI), we further characterized spatial aspects of print tuning in the 5th grade. The robust N1 tuning deficit for print in the dyslexic 2nd graders had largely disappeared by grade 5 consistent with a developmental delay. Reduced word-specific activation in dyslexic 5th grader's fMRI data occurred bilaterally in middle temporal regions and in the left posterior superior sulcus. Although no group differences in inferior occipito-temporal regions appeared in the whole brain analysis, a region of interest analysis of the Visual Word Form Area revealed that control children showed a more lateralized word-specific activation pattern than the children with dyslexia. The results suggest that while impaired N1 tuning for print plays a major role for dyslexia at the beginning of learning to read, other aspects of visual word form processing in the same region remain impaired in dyslexic children after several years of reading practice. Overall, neural deficits associated with dyslexia appear to be plastic and to change throughout development and reading acquisition.

Introduction

Reading skills are based on the ability to rapidly recognize visual language code. There is growing evidence that impaired visual expertise for print also plays a role for the development of dyslexia (Helenius et al., 1999, Paulesu et al., 2001, Maurer et al., 2007).

In adult skilled readers the N1 (or N170) component that peaks between about 150 and 200 ms in the visual event-related potential (ERP) has been shown to index visual expertise for print, as it is typically larger for words than for control stimuli such as symbol strings (Bentin et al., 1999, Maurer et al., 2005a, Maurer et al., 2005b). Intracranial recordings (Nobre et al., 1998) and source estimations of ERP and MEG data (Tarkiainen et al., 1999, Maurer et al., 2005b) suggest that print-specific activation occurs in the inferior parts of the occipito-temporal cortex.

This is in agreement with functional magnetic resonance imaging (fMRI) studies that have identified a “Visual Word Form Area” (VWFA) in the left mid-fusiform gyrus region that shows sensitivity to word forms (Cohen et al., 2000). More recently the VWFA was shown to lie within the anterior part of an extended visual word form system (VWFS) located in the inferior occipito-temporal cortex that shows a posterior-to-anterior gradient of word form specificity (Brem et al., 2006, Vinckier et al., 2007, van der Mark et al., 2009). This VWFS has been postulated to be tuned for print and to form a hierarchical system of local combination detectors with sensitivity to larger fragments of words increasing in more anterior regions (Dehaene et al., 2005).

Additional empirical evidence that this visual expertise is established through reading training and grapheme–phoneme mapping comes from developmental longitudinal and cross-linguistic studies. Developmental studies showed that the N1 tuning for print develops with learning to read (Maurer et al., 2006, Parviainen et al., 2006, Brem et al., 2010). Whereas the N1 tuning for print was absent in non-reading kindergarten children who could already categorize letters, it had developed strongly in the same children after the first 1.5 years of reading training in school (Maurer et al., 2006). Print tuning also emerged after a short grapheme–phoneme training in preschoolers (Brem et al., 2010), suggesting that it starts to develop at the very beginning of learning to read. The size of the N1 print tuning seems to follow an inverted U-shape during acquisition of reading skills, with little or no tuning at the onset of reading training in kindergarten, with large, coarse tuning in novice readers in 2nd grade, and with more subtle, fine-tuning in expert adult readers (Maurer et al., 2006), in agreement with the development of other forms of perceptual expertise (Palmeri and Gauthier, 2004).

Expertise as the driving factor behind the N1 specialization for print is also suggested by cross-linguistic studies (Wong et al., 2005, Maurer et al., 2008). Native English speakers without any knowledge of Japanese showed a reduced N1 over left occipito-temporal electrodes in response to Japanese scripts compared to native Japanese speakers (Maurer et al., 2008). This result suggests that the Japanese speakers' experience with Japanese scripts led to tuning of fast visual processes for these stimuli.

Reduced print tuning may also be characteristic for dyslexic reading. In agreement with the phonological deficit hypothesis (Bradley and Bryant, 1978, Ramus et al., 2003), brain imaging studies have revealed reduced temporo-parietal activation in phonological tasks that was associated with dyslexia (Rumsey et al., 1997, Shaywitz et al., 1998, Temple et al., 2001, Hoeft et al., 2007). A robust finding in such dyslexia studies, however, was also an underactivation in inferior occipito-temporal regions when the tasks involved explicit or implicit reading of words or pseudowords (Shaywitz et al., 1998, Paulesu et al., 2001; see Richlan et al., 2009, for a recent meta-analysis). Similarly, an MEG study in adults with severe dyslexia revealed that the fast print tuning that occurred in normal readers within 200 ms after stimulus presentation, was reduced in dyslexic readers (Helenius et al., 1999). This study left open the question whether the impaired print specialization in dyslexia was due to life-long reading impairment and thus developing gradually — maybe even as late as in adulthood, or whether this impairment also played a role during reading acquisition in school children.

In a recent longitudinal study (Maurer et al., 2007), following children from kindergarten to 2nd grade, we demonstrated that the latter was the case: children who developed dyslexia by grade 2 showed an impaired development of print tuning during the first 1.5 years of reading training in school. An open question is, whether this tuning impairment remains throughout school or whether it gets reduced similar to the reduction of print tuning during normal development (Maurer et al., 2006). To answer this question in the present paper, we followed up the same children from our longitudinal study, when they were in 5th grade, and presented them the same experiment for assessing print tuning as reported previously (Maurer et al., 2005b). We compared children who were dyslexic at both time points to children who were normally reading at both time points. Given the developmental trajectory of print specialization in normal readers (Maurer et al., 2006), we expected that the impairment in dyslexic children would become smaller in 5th grade reflecting a developmental delay of print tuning in dyslexia. Based on our previous results, we focused on the word–symbol comparison in the P1 and N1 components which peak within the first 250 ms of children's ERPs.

Our recent fMRI study had shown decreased print specialization in dyslexic 4th and 5th graders' visual word form system with an explicit reading task (van der Mark et al., 2009). Using the same task as for the ERP data in 2nd and 5th grade we also obtained fMRI data in 5th grade. We wanted to test whether impaired print specialization in dyslexia could be localized to inferior occipito-temporal regions, particularly to the VWFA (McCandliss et al., 2003) even when the test required reading only implicitly (Brunswick et al., 1999), and how VWFA tuning deficits would relate to fast print tuning deficits, as measured in the ERP data.